A welded assembly is not a machined part with a seam in it. The act of welding changes the shape of the assembly, and every surface that was cut or machined before the weld has moved by the time the part cools. Machining after welding exists to recover that lost geometry, and it works only when the shop understands what moved and why.
The movement is not random. Weld metal shrinks as it solidifies, the surrounding plate contracts as it cools, and the assembly pulls toward the joint. Thin sections pull more than thick ones. A joint welded on one side pulls more than the same joint welded on both. The result is a part that is close to the drawing in some places, out of square in others, and slightly bowed between them.
Machining starts from a datum chosen on a drawing, and a welder starts from whatever surface the fixture offered. Those are rarely the same surface. If the datum face is a plate that has bowed during welding, every dimension referenced to it inherits the bow, and the part is dimensionally correct and physically wrong at the same time.
The usual answer is to machine a datums after welding: face one surface flat, then use it to locate everything else. That single decision fixes more weldment problems than any other, because it converts an uncertain reference into a known one. It also means the assembly needs enough stock on that face to clean up, which is a drawing decision made long before anyone stands at the mill.
A flange that has been welded to a tube rarely comes out flat. The weld contracts around the circumference unevenly, the tube end is not perfectly square, and the flange itself may have distorted from the heat. Facing removes the high spots and produces a surface that will seal or bolt down.
The trap is that facing a flange removes material from one side only. If the flange is thin, the cut can relieve the residual stress in the plate and the flange will bow again after it comes off the table. A part that was flat in the vice and is dished on the bench is a common and expensive discovery. Heavy flanges behave better, and so does a flange that was stress relieved before the finishing cut.
Parallelism matters as much as flatness. A flange face that is flat but not parallel to the opposite face will bolt down and leak, or will load the joint unevenly. Both faces usually have to be machined in the same setup, or in two setups with the part turned against a proven face.
Tubes welded into a plate, bosses welded onto a casting, and housings assembled from two halves all produce bores that are no longer round. The weld pulls the bore in on the side nearest the joint, and the resulting ovality is enough to stop a bearing, a valve or a spigot from seating.
Boring recuts the bore on its own axis, which is the correct repair only if that axis is still where the drawing says. On a distorted part, the bore may be round and correct in diameter while the axis has shifted, and a bore that is round but displaced is worse than one that is oval and centred. That is why boring is usually set up from the mating face or from the opposite bore rather than from the damaged one.
Reaming removes a small amount of material from a hole that is already close, and its purpose is a combination of diameter control and surface finish. A drilled hole has a helix in it and a slightly irregular wall; a reamed hole is straight, round and smooth enough for a pin or a shaft to bear on.
Reaming will not correct position. If the hole is in the wrong place, reaming makes a smooth hole in the wrong place. It also will not fix a hole that has been work hardened by a blunt drill, because the reamer will rub rather than cut and the finish will be worse than the hole it started in. Reaming is a finishing operation and should be fed a hole that is already sound.
The material a tool meets after welding is not uniform. Beside the fusion line the alloy has been through a thermal cycle, and in stainless that can mean a harder, more abrasive band. In titanium it can mean a surface layer that is harder and more brittle than the parent metal. In aluminium the weld and the surrounding metal may be a different temper altogether.
What that means in practice is that a cut which starts easily can change character halfway through, and a tool set for the parent metal may be wrong for the band. It also means the finish can vary along the same bore. A machinist who notices a change in sound or chip form partway through a pass is reading exactly this, and the answer is usually a change of feed rather than more pressure.
The titanium cases are the ones to treat with most care, because a cut that smears or burns the surface leaves damage that is hard to see and hard to remove. The parts we machine and weld on that side are illustrated under TIG titanium welding parts.
Machining a weldment often means cutting across a seam, a tack or a repair. An interrupted cut impacts the tool once per revolution, and each impact is a small shock that dulls the edge faster than continuous cutting. On a thin-walled assembly it also sets the part ringing, which shows up as chatter on the surface and as a dimension that varies around the bore.
Support is the usual remedy. Packing a thin tube, clamping close to the cut, or machining with the part still bolted to its fixture all reduce the ringing. Where the geometry allows, a slower approach with a lighter cut leaves a better surface than a heavy pass that the part cannot absorb quietly.
Most welded assemblies are machined twice. The pre-weld machining brings the parts to rough size, provides the features the fixture locates on, and leaves stock where the finishing cut will remove distortion. The post-weld machining brings the critical features to final size once the assembly has stopped moving.
Doing all the machining before the weld is cheaper and faster and is the right answer for parts where the welded features are not critical. Doing all of it after the weld is sometimes unavoidable and is usually the most expensive option, because the part has to be set up again from geometry that has already moved. The decision belongs on the drawing as a note about which features are machined after welding, not as an assumption in the shop.
Machining after welding is a repair of a predictable distortion, so the useful questions are about what the part has to do once it is finished.
Where machining follows welding in our own work, the parts are handled as assemblies rather than as loose pieces — the fabricated side is shown under titanium fabrication components, the stainless assemblies under stainless steel exhaust system, and the welding practice behind both under TIG stainless steel welding and TIG aluminum welding. How a job is quoted and sequenced from a drawing or a sample is described on our customization page.
